Gaia Data Release 1: Catalogue validation

11.2 MB·

by  F. Arenou

Astronomy&Astrophysicsmanuscriptno.GAIA-CS-CP-OPM-FA-072 (cid:13)cESO2017
January3,2017
Gaia Data Release 1
Catalogue validation
F.Arenou1,X.Luri2,C.Babusiaux1,C.Fabricius2,A.Helmi3,A.C.Robin4,A.Vallenari5,S.Blanco-Cuaresma6,
T.Cantat-Gaudin5,K.Findeisen1,C.Reylé4,L.Ruiz-Dern1,R.Sordo5,C.Turon1,N.A.Walton7,I-C.Shih1,
E.Antiche2,C.Barache8,M.Barros9,M.Breddels3,J.M.Carrasco2,G.Costigan10,S.Diakité4,L.Eyer6,F.Figueras2,
L.Galluccio11,J.Heu1,C.Jordi2,A.Krone-Martins9,R.Lallement1,S.Lambert8,N.Leclerc1,P.M.Marrese12,13,
A.Moitinho9,R.Mor2,M.Romero-Gómez2,P.Sartoretti1,S.Soria2,C.Soubiran14,J.Souchay8,J.Veljanoski3,H.
Ziaeepour4,G.Giuffrida13,E.Pancino15,A.Bragaglia16
7 1 GEPI,ObservatoiredeParis,PSLResearchUniversity,CNRS,Univ.ParisDiderot,SorbonneParisCité,5PlaceJulesJanssen,
1
92190Meudon,France
0
e-mail:[email protected]
2 2 InstitutdeCiènciesdelCosmos,UniversitatdeBarcelona(IEEC-UB),MartíFranquès1,E-08028Barcelona,Spain
n 3 KapteynAstronomicalInstitute,UniversityofGroningen,Landleven12,9747ADGroningen,TheNetherlands
a 4 InstitutUTINAM,CNRS,OSUTHETAFranche-ComtéBourgogne,Univ.BourgogneFranche-Comté,25000Besançon,France
J 5 INAF,OsservatorioAstronomicodiPadova,VicoloOsservatorio,Padova,I-35131,Italy
6 ObservatoiredeGenève,UniversitédeGenève,CH-1290Versoix,Switzerland
1
7 InstituteofAstronomy,UniversityofCambridge,MadingleyRoad,CambridgeCB30HA,UnitedKingdom

8 SYRTE,ObservatoiredeParis,PSLResearchUniversity,CNRS,SorbonneUniversités,UPMCUniv.Paris06,LNE,61avenue
]
M del’Observatoire,75014Paris,France
9 CENTRA,UniversidadedeLisboa,FCUL,CampoGrande,Edif.C8,1749-016Lisboa,Portugal
I 10 LeidenObservatory,LeidenUniversity,NielsBohrweg2,2333CALeiden,TheNetherlands
.
h 11 LaboratoireLagrange,Univ.NiceSophia-Antipolis,ObservatoiredelaCôted’Azur,CNRS,CS34229,06304Nicecedex,France
p 12 INAF-OsservatorioAstronomicodiRoma,ViadiFrascati33,00078MontePorzioCatone(Roma),Italy
- 13 ASIScienceDataCenter,ViadelPolitecnico,Roma
o 14 Laboratoired’astrophysiquedeBordeaux,Univ.deBordeaux,CNRS,B18N,alléeGeoffroySaint-Hilaire,33615Pessac,France
r 15 INAF-OsservatorioAstrofisicodiArcetri,LargoEnricoFermi5,I-50125Firenze,Italy
t
s 16 INAF-OsservatorioAstronomicodiBologna,viaRanzani1,40127Bologna,Italy
a
[

1
v ABSTRACT
2
9 Context. Before the publication of the Gaia Catalogue, the contents of the first data release have undergone multiple dedicated
2 validationtests.
0 Aims.Thesetestsaimatanalysingin-depththeCataloguecontenttodetectanomalies,individualproblemsinspecificobjectsorin
0 overallstatisticalproperties,eithertofilterthembeforethepublicrelease,ortodescribethedifferentcaveatsofthereleaseforan
. optimalexploitationofthedata.
1
Methods.DedicatedmethodsusingeitherGaiainternaldata,externalcataloguesormodelshavebeendevelopedforthevalidation
0
processes. They are testing normal stars as well as various populations like open or globular clusters, double stars, variable stars,
7
quasars. Properties of coverage, accuracy and precision of the data are provided by the numerous tests presented here and jointly
1
analysedtoassessthedatareleasecontent.
:
v Results.Thisindependentvalidationconfirmsthequalityofthepublisheddata,GaiaDR1beingthemostpreciseall-skyastrometric
i andphotometriccatalogueto-date.However,severallimitationsintermsofcompleteness,astrometricandphotometricqualityare
X
identifiedanddescribed.Figuresdescribingtherelevantpropertiesofthereleaseareshownandthetestingactivitiescarriedoutvali-
r datingtheuserinterfacesarealsodescribed.Aparticularemphasisismadeonthestatisticaluseofthedatainscientificexploitation.
a
Keywords. astrometry–parallaxes–propermotions–methods:dataanalysis–Surveys–Catalogs–
1. Introduction astronomicalprojects,initiatedinthe late19th centurywiththe
Carte du Ciel (Jones 2000), and a direct successor of the ESA
Hipparcosmission(Perrymanetal.1997).
Thispaperdescribesthevalidationofthefirstdatareleasefrom
the European Space Agency mission Gaia (Gaia Collaboration Despite the precautions taken during the acquisition of the
et al. 2016b). In a historical perspective, Gaia, following in the satelliteobservationsandwhenbuildingthedataprocessingsys-
footsteps of the great astronomical catalogues since the first by tem, it is a difficult task to ensure perfect astrometric, photo-
HipparchusofNicaea,describesthestateoftheskyatthebegin- metric, spectroscopic and classification data for a one billion
ningofthe21stcentury.Itistheheirofthemassiveinternational source catalogue built from the intricate combination of many
Articlenumber,page1of34
A&Aproofs:manuscriptno.GAIA-CS-CP-OPM-FA-072
data items for each entry. However, several actions have been secondarysources,containsthepositionsandGmagnitudesfor
undertakentoensurethequalityoftheGaiaCataloguethrough 1140622719 sources brighter than about magnitude G = 21.
bothinternalandexternaldatavalidationprocessesbeforeeach Anannexofvariablestarslocatedaroundthesoutheclipticpole
release. The results from the external validation work are de- isalsopartofthereleasethankstothelargenumberofobserva-
scribedinthispaper. tionsmadeduringtheEPSLmode.
TheGaiaDR1: ThereisanexhaustivedescriptionoftheGaia The Catalogue Validation: In terms of scientific project, the
operationsandinstrumentsinGaiaCollaborationetal.(2016b), quality of the released data has been controlled by two com-
oftheGaiaprocessinginGaiaCollaborationetal.(2016a)and plementaryapproaches:theverificationsdoneinternallyateach
the astrometric and photometric pre-processing is also detailed stepoftheprocessingdevelopmentinordertoanswertheques-
inFabriciusetal.(2016).Forthisreasonwementionhereonly tion: are we building the Catalogue correctly? and the valida-
what is strictly necessary and invite the reader to refer to the tionsattheend:isthefinalCataloguecorrect?
abovepapersortotheGaiadocumentationfordetails. Itisfundamentaltonotethatthefirststepofthevalidations
The Gaia satellite is slowly spinning and measures the is logically represented by the many tests implemented in the
fluxes and observation times of all sources crossing the focal GaiaDPACgroupsbeforeproducingtheirowndata,andwhich
plane (their Gaia transit), sending to the ground small win- aredescribedindedicatedpublications,Lindegrenetal.(2016)
dows of pixels around the sources. These times correspond for the astrometry, Evans et al. (2016) for the photometry, and
to one-dimensional, along-scan positions (AL in what follows) Eyeretal.(2016)forthevariability.
which are used in an astrometric global iterative solution pro- ToassesstheCataloguepropertiesandasafinalcheckbefore
cess (AGIS, Lindegren et al. 2016) which also needs to simul- publication,theDPACdeemedusefultoimplementasecondand
taneously calibrate the instruments and reconstruct the attitude laststep:avalidationoftheCatalogueasawholeandactually,
of the satellite. A star crossing the focal plane is measured on thismustbestressed,afullyindependentvalidation.
9 CCDs in the astrometric instrument so the number of obser- TheactualCataloguevalidationoperationsbeganafterdata
vationsofastarcanbeupto9timesthenumberofitstransits. from the DPAC groups had been collected and a consolidated
On-board resources are able to cope with various stellar densi- Cataloguehadbeenbuiltbeforepublication.Atthisstep,nore-
ties; however, for very dense fields above 400000 sources per run of the data processing was possible, only the rejection of
squaredegree,thebrightersourcesarepreferentiallyselected. some stars (if strictly needed) and some cosmetic changes on
the data fields could be done. After the rejection of problem-
The photometric instrument is composed of two prisms, a
aticstars,aprocesslabelledasfiltering,thevalidationwasagain
BluePhotometer(BP)andaRedPhotometer(RP).Thiscolour
performed,andmostofthecataloguepropertiesdescribedinthis
informationisnotpresentintheGaiaDR1,onlytheG-bandpho-
paperrefertothispost-filtering,published,finalGaiaDR1data.
tometry,derivedfromthefluxesmeasuredintheastrometricin-
strument being given. The CCD dynamic range does not allow
toobserveallsourcesfromthebrightestuptoG ∼ 21:sources The organisation of this paper is as follows: Sect. 2 sum-
brighter than G ∼ 12 would be saturated. To avoid this, Time marisesthedataandmodelsused.Section3describestheerro-
Delay Integration (TDI) gates are present on the CCD and can neous or duplicate entries found and partly removed. The main
be activated for bright sources, which in practice reduce their properties of the Gaia DR1 Catalogue are discussed, Sect. 4,
integrationtime(butalsocomplicatestheircalibration). fortheskycoverageandcompleteness,withamultidimensional
analysisinSect.5,theastrometricqualityofGaiaDR1inSect.6
Astrometry and photometry are then derived on-ground in
andthephotometricpropertiesinSect.7.Asaconclusion,rec-
independentpipelines,whicharepartoftheworkdevelopedun-
ommendationsfordatausagearegiveninSect.8.Thevalidation
dertheresponsibilityofthebodyinchargeofthedataprocess-
proceduresemployedintestingthedesignandinterfacesofthe
ingfortheGaiamission,theGaiaDataProcessingandAnalysis
archive systems are described in Appendix together with some
Consortium(DPAC,GaiaCollaborationetal.2016a).
illustrationsofthestatisticalpropertiesoftheCatalogue.
This first data release contains preliminary results based on
observations collected during the first 14 months of mission
since the start of nominal operations in July 2014. At the start
2. Dataandmodels
ofnominaloperationsofthespacecrafton25July2014,aspe-
cialscanninglawwasfollowed,theEclipticPoleScanningLaw 2.1. Dataused
(EPSL). In EPSL mode, the spin axis of the spacecraft always
2.1.1. Gaiadata
lies in the ecliptic plane, such that the field-of-view directions
pass the north and south ecliptic poles on each six-hour spin. Twomonthsbeforethefinalgo-aheadtopublishtheGaiaDR1
ThenfollowedtheNominalScanningLaw(NSL)withapreces- Catalogue,wereceivedtheofficialpreliminaryCatalogue,called
sionrateof5.8revolutionsperyear,startingon22August2014. pre-DR1 in what follows, which was validated, then subse-
Aswewillnoticebelow,theEPSLmodeleftsomeimprintson quentlyfiltered,asdescribedinSect.3,toproducetheGaiaDR1
theCataloguecontentandscientificresults. Catalogue.Generallyspeaking,thevalidationworkhashadac-
Gaia DR1 contains a total of 1 142 679 769 sources, the cess to the same fields as published in Gaia DR1 so that any
astrometric part of Gaia DR1 being built in two parts: the pri- user can reproduce the work indicated below. For example we
mary sources contains positions, parallaxes, and mean proper did not have access to any individual transit data or calibra-
motionsfor2057050ofthestarsbrighterthanaboutmagnitude tiondata,ormoregenerallytothemainGaiadatabase,andthis
V = 11.5 (about 80% of these stars). This data set, the Tycho fostered developing methods independent from the work done
Gaia Astrometric Solution (TGAS), was obtained through the within the Gaia groups producing the data. A few supplemen-
combination of the Gaia observations with the positions of the tary fields were however kindly made available for validation
sources obtained by Hipparcos (ESA 1997) when available, or purposes, such as the preliminaryG andG magnitudes (in
BP RP
Tycho-2 (Høg et al. 2000b). The second part of Gaia DR1, the ordertostudypossiblechromaticeffects).
Articlenumber,page2of34
F.Arenouetal.:GaiaDataRelease1–Cataloguevalidation
2.1.2. SimulatedGaiadata All data integrity issues were fixed before the data release.
ForTGASsolutionswealsocheckedindividualvaluesofproper
In the course of the preparation of the data validation, we also motionsandparallaxlookingfore.g.negativeparallaxesorunre-
needed simulated data, mostly for testing the astrometry of the alistictangentialvelocities.Wethencheckedtheuncertaintiesof
TGAS solution. For this purpose we built a simulated cata- thefiveastrometricparameterstomakesurethattheydecreased
logue, called Simu-AGISLab in what follows, which contained withthenumberofobservations,ortoseeiftherewereHealpix
astrometric data for the Tycho-2 stars, on top of which were pixelswithanunusuallyhighfractionoflargeuncertainties.All
addedsimulatedTGASastrometricerrors.Simu-AGISLabused inallwewereparticularlyinterestedinregionsontheskywhere
as simulated proper motions the Tycho-2 ones, but they were dubious values occur with higher frequency than in typical ar-
“deconvolved” using the formula indicated in Arenou & Luri eas, with the aim of excluding if needed such regions from the
(1999, Eq. 10) to avoid a spurious increase of their dispersion release. Although some poorly scanned regions were identified
withtheTGASastrometricerrorsaddedbythesimulation.The asproblematic,nonewerefinallyexcluded.
simulatedparallaxeswereaweightedaverageof“deconvolved”
Sources brighter than about 12 mag are observed with
Hipparcosparallaxes(fornearbystars)andthephotometricpar-
“gates”, i.e. with reduced exposure time. We therefore checked
allaxesfromthePickles&Depagne(2011)catalogue(formore
that the astrometric standard uncertainties did not show rapid
distantstars).ThesimulatedTGASastrometricerrorswerepro-
changesasafunctionofmagnitude.
ducedasdescribedintheTycho-GaiaAstrometricSolutiondoc-
We found only a few minor issues in the Gaia DR1 as-
ument (Michalik et al. 2015), based on solution algorithms de-
trometry as for the data ranges. Large values of fields like
scribedinLindegrenetal.(2012,Sect.7.2).
astrometric_excess_noise1 and astrometric_excess-
In addition, global simulations of the Gaia data generated _noise_sig that statistically were expected for only about
by the DPAC group devoted to this purpose were also used for a thousand sources are actually present in about 205 million
validationtaskscomparingmodelswithdata(seeSect.2.3). sources, including nearly the entire TGAS sample. These large
values reflect the large errors introduced by the preliminary at-
titude solution for the Gaia spacecraft; a better solution will be
2.1.3. Externaldata
usedinfuturereleases(Lindegrenetal.2016)andweexpectthis
The comparison of Gaia DR1 to external catalogues is a tricky problemwillbesolved.Inaddition,4288sourceshavepositions
taskastheGaiaCatalogueisuniqueinmanyways:itcombines basedononlytwoone-dimensionalmeasurements,providingan
the angular resolution of the Hubble Space Telescope with a astrometric solution with no degrees of freedom. These mini-
complete survey all over the sky in optical wavelength, down mally constrained solutions are expected to go away as more
toaG-magnitude(cid:39)21,unprecedentedastrometricaccuracyand dataarecollected.
all-skyhomogeneousphotometricdata. Wetestedwhethersourceshadenoughastrometricmeasure-
ments to allow for a 2- or 5-parameter solution, as appropriate.
However, the comparison with external catalogues is one
Wethencomparedthedistributionofastrometricgoodness-of-fit
waytowardsadeeperunderstandingofmanyoftheparameters
indicatorswiththeirexpecteddistributions.
describingtheperformanceoftheCatalogue:overallskycover-
age, spatial resolution, catalogue completeness and, of course, Photometry and astrometry were derived in independent
precisionandaccuracyofthedifferenttypesofdataforthevari- pipelineseachofwhichcoulddecidetorejectordownweighta
ouscategoriesofobjectsobservedbyGaia.BesidestheHippar- numberofindividualobservationsforagivensource.Wethere-
cos and Tycho-2 catalogues, many other catalogues have been forecheckedifthenumberofvalidobservationswassimilarin
used, especially chosen for each of these tests. They are de- thetwopipelines.Ifmorethanhalfoftheobservationswerere-
scribedineachoftherelevantsubsections. jected,andifthenumberofvalidobservationsineachpipeline
adds up to less than the total number of observations for the
Thecross-matchbetweenTGASandtheexternalcatalogues
source, there is a problem: it is not possible to know if the as-
or compilations has been done using directly Tycho-2 or Hip-
trometricandphotometricresultsrefertothesameobjectore.g.
parcosidentifiers,eitherprovidedinthepublicationsorobtained
to different components of a binary star. This problem affects
throughSIMBADqueries(Wengeretal.2000)usingtheidenti-
lessthan9000sourcesinGaiaDR1andweexpectittobealso
fiersgivenintheoriginalpapers.ForthefullGaiaDR1tests,a
solvedinfuturereleases2.
positionalcross-matchhasbeenused.
2.3. Galaxymodels
2.2. Dataintegrityandconsistency
Modelscontainasummaryofourpresentknowledgeaboutthe
GaiaDR1isthecombinedworkofhundredsofpeopledivided
stars in the Milky Way. This knowledge is obviously imperfect
intodozensofgroupsworkingonseveralcomplementaryyetin-
andoneexpectsthatmanyofthediscrepanciesbetweenmodels
dependent pipelines. In addition to testing the data themselves,
andrealGaiadatatobeduetothemodelsthemselves.However,
therefore, we tested the data representations to ensure that all
atthelevelofourcurrentknowledge,ifamodelperformswith
catalogueentrieswerevalidandself-consistent.Wecheckedthat
asatisfactoryaccuracycomparedtoexistingdata,itcanbeused
cataloguevalueswerefinite,thatdatawerepresent(ormissing)
forGaiavalidation(atthelevelofthisaccuracy).Thisiswhatwe
whenexpected,thatallfieldswereintheirexpectedranges,that
have done in the set of tests based on models. These tests may
observation counts agreed with each other, that source identi-
fiers were unique, that correlation coefficients formed a valid
1 Roughly speaking, this is the noise which should be added to the
correlationmatrix,thatfluxesandmagnitudeswererelatedasex-
uncertaintyoftheobservationstoobtainaperfectfitfortheastromet-
pected,thatthepositionsobtainedfromtheequatorial,ecliptic, ric model. The fields of the Gaia Catalogue are described at https:
and galactic coordinates agreed, and so on. We also confirmed //gaia.esac.esa.int/documentation/GDR1/datamodel/
thattheGaiaDR1indifferentdataformatsindeedcontainedthe 2 Thesestarsarenotflagged,butcanbefoundusingphot_g_n_obs,
samedata. astrometric_n_good_obs_al,matched_observations
Articlenumber,page3of34
A&Aproofs:manuscriptno.GAIA-CS-CP-OPM-FA-072
supersedethevalidationusingexternaldatainregionsofthesky
wheredataaretooscarce,orinmagnituderangeswhereexisting
data are not accurate enough or incomplete, or in case they do
notexistinlargeportionsofsky(suchase.g.parallaxes).
OnGaiaDR1,threekindsoftestshavebeenperformed:tests
onstellardensities,testsonpropermotions,andtestsonparal-
laxes.Inalltestsweanalysedthedistributionontheskyofthe
modeldensitiesandofthestatisticaldistributionofastrometric
parameters(propermotionsandparallaxes)andcomparedthem
with Gaia data. In order to establish a threshold for test results
wecomparedthemodelwithpreviouscataloguesonportionsof Fig.1.HistogramofG magnitudesforTGASstars(a)beforeand(b)
skywhenavailable.ForthisfirstdatareleaseonlytheBesançon aftervalidationfiltering.
Galactic Model (Robin et al. 2003) has been used for compar-
isonswithGaiadata.
3.1.2. Dataafterfiltering
Figure1bshowstheresultingmagnitudedistributionforTGAS
in Gaia DR1, i.e. after full filtering. There is a remaining tail
3. Erroneousorduplicateentries with352sourcesfainterthanG =13.5mag,andthepresenceof
such sources in TGAS calls for an explanation. We have taken
The pre-DR1 Catalogue received for validation was subject to
acloserlookatthe60faintestTGASstarsofwhichthebright-
several tests concerning possible erroneous entries. This led est hasG = 14.98mag. Of these 60 stars, 25 have a neighbour
to the filtering of a significant number of sources (37433092 brighter than G = 13.5mag and closer than 5(cid:48)(cid:48) in Gaia DR1
sourceswereremoved,3.2%oftheinputsources).Asthisfilter-
suggestingthatthewrongstarmayhavebeenusedintheTGAS
ing was obviously not perfect (removing actual sources while
solution,whichisthereforenotvalid.Oftheremaining35stars,
conservingerroneousones),andhadanimpactontheCatalogue justoverhalf(18)havefromonetofourneighbourswithin5(cid:48)(cid:48).
content,therationale,methodsusedandresultsaredescribedin
In these cases we may be dealing with spurious Tycho-2 stars.
thissection.
Tycho-2 (Høg et al. 2000a) was using an input star list domi-
nated by photographic catalogues, and a blend of sources may
therefore have been seen as a single bright source. It may then
3.1. ErroneousfaintTGASsources
happenthataTycho-2solutionwasderivedfromthemixedsig-
nalofcontaminatingsources.Weseethatasalikelyexplanation
3.1.1. Databeforefiltering
formostofthesecases.ForstarsthatareisolatedinGaiaDR1,
As can be seen in Fig. 1a, there was a significant number of spurious Tycho-2 stars cannot be excluded, but in at least one
objects (2381 sources) in the pre-DR1 version of TGAS that case,thefaintGaiasourceturnsouttobeavariableoftheRCrB
hadG (cid:38) 14mag,i.e.clearlyfainterthanwhatwasexpectedfor type.Thisstar(HIP92207)hasG =16.57maginGaiaDR1,but
Tycho-2.ThisledtothestudyoftheGphotometryforthesestars isasbrightasVT =10.29maginTycho-2.Thisisingoodagree-
and,beyond,forthewholecatalogue. mentwithavailablelightcurves.Itistooearlytosayifthereare
morehighamplitudevariablesinthesample.
Aparticularconcernhasbeentocatchcoarseprocessinger-
rorsinthephotometry.Forbrightsources,theexposuretimein
each CCD on-board Gaia is reduced by activating special TDI 3.2. Duplicateentries
gates on the device as the star image crosses the CCD. This
smallerexposuretimeisthentakenintoaccountwhencomput- 3.2.1. GaiaDR1beforefiltering
ing the flux. However, in some rare occasions the information
Before launch, a catalogue with known optical astrometric and
on gate activation did not reach the photometric pipeline. The photometricinformationofsourcesuptomagnitudeG =21had
resultwasartificiallylowfluxesinthatparticulartransit,andfor
beenbuiltinordertobeusedasInitialGaiaSourceList(IGSL,
reasonsbeyondthescopeofthispaper,thiscouldupsetthepro-
Smart&Nicastro2014).
cessingandleadtoerroneousGmagnitudes.
Stars from IGSL may have initially contained duplicates
Wethereforespecificallycheckedifsourcesappearedmuch originating from e.g. overlapping plates. Automatically gen-
fainterinG thaninbothGBP andGRP,thepreliminaryversions erated catalogues such as Gaia DR1 may also have multiple
of photometry to be published in later releases (Riello et al. copies of a source for a variety of reasons, including poor
2016).Inpracticethelimitwassetat3maginordernottoelim- cross-matching of multiple observations, inconsistent handling
inatediffuseobjectswithabrightcore,e.g.galaxies,whichwere ofclosedoubles,orotherobservationalorprocessingproblems,
expected to be bright in the diaphragm photometry ofGBP and besidetheduplicatesoriginatingfromtheIGSL.Totestfordu-
GRP; stars with G −GBP > 3 and G −GRP > 3, thus where a plicatesourceswecross-matchedtheGaiacatalogueagainstit-
problemwithGwassuspected,werefiltered(164446TGASor self,identifyingpairsofsourcesthatcouldnotpossiblybereal
secondarysources). doubles, either because they fell within one pixel (59 mas) of
WhilethemediannumberofG-bandobservationspersource each other or because their positions were consistent to within
is72inGaiaDR1,itwasalsofoundthatroughlyhalfofthetoo 5σ. Only reference epoch positions were used, with no correc-
faint TGAS sources had fewer than 10 CCD observations, and tionsforhighpropermotionstars.
indeed,onthewholecataloguestarswithlessthan10observa- It was found that the pre-DR1 Gaia catalogue contained
tions clearly behaved incorrectly. This led to the removal of all 71 million sources with a counterpart within one pixel or 5σ.
sourceswithlessthan10Gobservationsfrompre-DR1(746292 Most appeared in pairs, but some were clustered in groups of
TGASorsecondarysources). up to eight duplicates. Up to one third of sources around G ∼
Articlenumber,page4of34
F.Arenouetal.:GaiaDataRelease1–Cataloguevalidation
estimatedbyafactor2forpositionsand4formagnitudes.While
thisresultcannotbeextrapolatedtoallnormal(notduplicated)
stars, this gives at least an upper limit and justifies in any case
thepresenceoftheduplicated_sourceflag.
AcomparisonwiththeWashingtonVisualDoubleStarCat-
alogue(WDS,Masonetal.2001)confirmsthatsomeduplicates
remain,ascanbeseenwiththeexcessofstarswithanearzero
separationinthebottomleftofFig.19b.
In high density fields, there is a chance to get several stars
very close to each other by chance only, i.e. optical doubles.
Trying to remove more duplicates would lead to removing ac-
tual stars by mistake. The adopted filtering may actually have
Fig. 2. Number of pairs of sources vs their angular separation in the beenareasonablecompromise,untiltheexpectedimprovement
field(l=350◦,b=0◦)before(red)andafterfiltering(green).Theline inGaiaDR2.
correspondstoarandomdistributionupto10(cid:48)(cid:48)ofthelatter.
4. SkycoverageandcompletenessofDR1
The Gaia DR1 release is expected to be incomplete in various
ways, full detail of these limitations being described in Linde-
grenetal.(2016);GaiaCollaborationetal.(2016a):
– Gaia DR1 is based on 14 months of data only. As a result,
someregions,especiallyatloweclipticlatitudes,havebeen
poorlyobserved,bothintermsofthenumberofobservations
andofthecoverageinscanningdirections,seeforexample
Fig. 2 of Gaia Collaboration et al. (2016a). Stars with less
Fig.3.Effectofduplicatestarsinafieldofradius4◦aroundtheSouth than5focalplanetransitshavebeenfilteredout;
pole:(a)originaldensitymapinpre-DR1beforevalidationfiltering,(b) – starswithalowqualityastrometrysolutionforwhateverrea-
duplicatesfound,(c)afterduplicatesfiltering. sonhavebeenfilteredout;
– brightstarsorhighpropermotionsstarsmaybemissing;
– faintstarsaremissinginverydenseareas(forstellardensi-
tieshigherthan∼400000starspersquaredegreeatG <20);
11mag were affected, far more than at much brighter or much
– stars with extremely blue or red colours have been filtered
faintermagnitudes.
outduringthephotometriccalibration.
For Gaia DR1, we removed all but one source from each
groupofclosematches,selectingthesourcewiththemorepre- Thetestspresentedinthissectionaimatabettercharacteri-
cise parallax (if present) and breaking ties by the source with sationoftheobjectcontentofDR1,includingTGAS,asforthe
moreobservations,followedbythebetterpositionorphotomet- homogeneity of the sky distribution and the small scale com-
ric error. Because duplicated sources may have compromised pleteness of the Catalogue. These tests have been performed
astrometry or photometry (e.g., if a source was duplicated be- from different points of view, for various populations and us-
causeofacross-matchingproblem),thesurvivingsourceswere ingvariousinputsandmethods:usingthecharacteristicsofGaia
markedwiththeduplicated_sourceflaginthefinalcatalogue data only (internal tests), using external data (all sky external
(35951041TGASorsecondarysources). catalogues,detailedcataloguesofspecificsamplesofstarsorof
Two examples of the effect of the filtering of duplicate specificregionsofthesky),orusingGalaxymodels.
sources are shown in Figs. 2 and 3. The result of the filtering
as done for Gaia DR1 is illustrated in Figs. 2 and 3c. The arte-
4.1. Limitingmagnitude
facts in Figs. 3a and 3b are the traces of the overlaps of photo-
graphicplatesusedinsomeofthesurveysfromwhichtheIGSL ThecompletenessofGaiaDR1istheresultofacomplexinter-
catalogue was built, causing an excess of duplicate sources in play between high stellar densities implying a possible overlap
GaiaDR1. oftheimagesonthefocalplane,scanninglawdefiningthenum-
beroftimesaregionwasobserved,anddataprocessing.Dueto
limited telemetry resources, the star images sent to ground fol-
3.2.2. GaiaDR1afterfiltering
lowed a decision algorithm which is a complex function of the
Although it is estimated that about 99% of the duplicates have magnitude.Inaddition,attheendofthedataprocessingafilter-
been removed, spurious sources may still remain in Gaia DR1. ingwasappliedtodiscardpoorsolutionsbothintheastrometry
Formal uncertainties on positions of these duplicates may have andinthephotometry.Asaresult,thedensitydistributionover
been underestimated, and the 5σ criterion on positional differ- theskyinthefinalCatalogueisnotasimplefunctionofthestel-
enceusedforrejectionmayfinallynothavebeenlargeenough. lardensity,asusuallyexpected.
This underestimation was suspected the following way: a pair A first, indirect information about the completeness is
made of one duplicate source and the source it duplicates ac- broughtbythelimitingmagnitudeoftheCatalogue.Skyvaria-
tually refers to one single source which dispatched part of its tionsofthe0.99quantileoftheGmagnitudeareshowninFig.4
observationsbetweenboth(dependingontheorientationofthe forTGASandthewholeCatalogue.Concerningthelatter,itap-
satellite scans). We used this property to compare the positions pearsthatGaiawilleasilyreachattheendofmission G >21in
andmagnitudesinpairsandfoundthatuncertaintieswereunder- asignificantfractionofthesky,evenifthisisstillverylimited
Articlenumber,page5of34
A&Aproofs:manuscriptno.GAIA-CS-CP-OPM-FA-072
reasonistheoccasionalsourceduplicationdescribedinSect.3,
which affects these magnitudes more. The loss is clearer for
stars brighter than 6mag, partly due to an insufficient number
ofbrightcalibrationsourcesforthebroadbandphotometers,so
no colour was available. TheG magnitude calibration includes
acolourterm(Carrascoetal.2016),soamissingcolourmeans
that no G-band photometry was produced, and the source did
notentertherelease.Starsbrighterthanabout5,andafraction
ofsourcesfainterthanthis,werealsoamongthesourcesnotkept
inTGASduetothebadqualityoftheirastrometricsolution.
TGAScompletenesshasalsobeentestedwithrespecttohigh
proper motion stars: a selection of 1098 high proper motion
(HPM)starshasbeenmadewithSIMBADonstarswithaTycho
orHIPidentifierandapropermotionlargerthan0.5arcsecyr−1
(proper motions mainly from Tycho-2 and Hipparcos). 40% of
this selection is not found in the TGAS solution, in particular
bright stars. All stars with a proper motion larger than 3.5 arc-
secyr−1areabsentfromTGAS.Starswithapropermotionlarger
than1arcsecyr−1 inTGAShavebeenconfirmedtohavealarge
propermotioninSIMBAD.
Fig. 4. Limiting magnitude: 99% percentile of the G distribution in
eclipticcoordinates:a)TGAS,b)fullCatalogue.
4.2.2. OverallskycoverageofGaiaDR1fromexternaldata.
TheoverallskycoverageofGaiaDR1hasbeentestedbycom-
parison with two deeper all sky catalogues: 2MASS (Skrutskie
et al. 2006) and UCAC4 (Zacharias et al. 2013). The tests per-
formed here use the crossmatch between Gaia DR1 and these
twocataloguesprovidedtotheusersintheGaiaArchive(Mar-
reseetal.2016).Thevariationovertheskyoffourkeyparame-
tersarechecked:thenumberofcross-matchedsources,themean
numberofneighbours(starswhichcouldhavebeenconsidered
ascross-matched,butforwhichthecross-matchwasnotasgood
Fig.5.SkydistributionofTycho-2starsnotinTGAS,ingalacticcoor- asfortheselectedsource=thebestneighbour),thenumberof
dinates. Gaiastarswiththesamebestneighbour,andthenumberofGaia
sourceswithoutanymatch.Finally,arandomsubsetofabout5
million sources has been selected in order to check, if any, the
different properties in magnitude, colour, proper motion, good-
for Gaia DR1; it seems however that one magnitude has been
nessoffit,etc...oftheabovefourcategoriesofstars.
lost in the under-scanned regions, and two magnitudes in the
Baadewindow.ThelimitingmagnitudeofTGASstarsalsohas
anamplitudeoftwomagnitudesoverthesky,withthebrightest
regions being also those with some astrometric deficiencies, as UCAC4. Only 5% of the UCAC4 catalogue does not have a
shownbelow. match in Gaia DR1. Their sky distribution (Fig. 6a) shows the
footprint of the Gaia scanning law. 7% of the UCAC4 sources
appear more than once in the cross-match table. We will refer
4.2. Overalllargescalecoverageandcompleteness to them as multiple-matches, it does not mean that this refer to
(or only to) duplicate Gaia entries as discussed in Sect. 3.2.1:
4.2.1. OverallskycoverageandcompletenessofTGAS
the Gaia resolution is much better than ground-based instru-
The overall TGAS content has been tested with respect to the mentssothatmultipleobjectsmayappearwhereground-based
Tycho-2 (Høg et al. 2000b) and Hipparcos Catalogues (Perry- cataloguesseeoneobjectonly;thosemultiple-matchesaredis-
man et al. 1997; ESA 1997) for detection of possible duplicate tributedmainlyinhighdensityregion,asexpected,buttheirsky
entries and characterisation of missing entries. TGAS contains distributionalsoshowstheGaiascanninglawfootprint(Fig.6b).
79%oftheHipparcosand80%oftheTycho-2stars.Oneofthe 258605sourceswithG<14appearintheGaiacataloguebutnot
reasonsforthemissingstarsisabadastrometricsolution,asall inUCAC4whichissupposedtobecompletetoaboutmagnitude
sources with a parallax uncertainty above 1 mas were not kept R = 16; their sky distribution (Fig. 6c) follows the Gaia scan-
in TGAS (validation tests done on preliminary data had indeed ninglawfootprintandrecallsthefootprintoftheTycho-2stars
shownseveralproblemsassociatedtothesestars).Theskydistri- notinTGAS(Fig.5).Adetailedinspectionofthosesourcesin-
butionoftheTycho-2sourcesnotpresentinTGASispresented dicates that a large portion of them are actually present in the
Fig.5,showingtheimpactoftheGaiascanninglaw(thenumber UCAC4 catalogue but that the cross-match could not be done,
ofobservationsandtheorientationofthescansbeingcorrelated the positional differences being beyond the astrometric uncer-
withthesolutionreliabilitycriteriafiltersappliedforGaiaDR1). tainties. This may be linked to the fact that a large portion of
ThedetailofthehistogramofFig.1showsthatstarsfainter those sources have been measured along uneven scan orienta-
than10.5maghavesufferedahigherlossthanaverage,alikely tions.
Articlenumber,page6of34
F.Arenouetal.:GaiaDataRelease1–Cataloguevalidation
Fig.6.SkydistributionversusUCAC4,ingalacticcoordinates;a)UCAC4sourcesnotinGaiaGaiaDR1(5%);b)UCAC4sourceswithmultiple
matchesinGaiaDR1;c)GaiaDR1sourceswithG<14notinUCAC4.
pared to those of missing galaxies (magnitudes, redshift, axis-
ratiosandradii).Unfortunately,only∼0.2%oftheSDSSgalax-
ies are present in Gaia DR1 due to the different filters applied.
Still some large resolved galaxies can have multiple detections
associatedtothem,tracingtheirshape.
4.2.3. CompletenessfromcomparisonwithaGalaxymodel
SinceGaiaDR1onlycontainsG magnitudesandpositions,the
validation with models consists in the comparison between the
distributionofstardensitiesovertheskyandarealisationofthe
Besançon Galactic Model (BGM, Robin et al. 2003), hereafter
version 18 of the Gaia Object Generator (GOG18, Luri et al.
2014). The simulation contains 2 billion stars including single
starsandmultiplesystems,andincorporatesamodelfortheex-
pectederrorsonGaiaphotometricandastrometricparameters.
In the validation process, star counts as a function of po-
Fig. 7. Sky distribution versus 2MASS, in galactic coordinates. a)
sitions and in magnitude bins have been compared with the
2MASS sources with J<14 not in Gaia DR1; b) 2MASS multiple- model (Fig. 8). Systematic differences in Galactic plane fields
matchesinGaiaDR1.
aremostlydueto3Dextinctionmodelproblems,butcouldalso
beduetootherinadequaciesofthemodel(suchaslocalclumps
not taken into account in a smooth model). These systematics
2MASS. For this test, we selected 2MASS stars with photo- are seeneven in brightmagnitude bins. On theother hand, dif-
metricqualityflagAAAandmagnitudeJ <14(thislimitcorre- ferencesatintermediatelatitudesintheregionoftheMagellanic
spondsroughlytoV < 20forA < 5).Asexpected,mostofthe Cloudsarenottobeconsideredbecausethesegalaxieshavenot
V
missingsourcesarelocatedinhighextinctionregionsalongthe been included in this GOG catalogue. There is no other strong
galacticplane,butsomeextrafeaturesarealsoapparentshowing difference between data and model that could warn about the
theGaiascanninglawfootprint(Fig.7a).The2MASSmultiple- quality of the data at magnitudes brighter than 16. However at
matcheshaveaskypattern(Fig.7b)similartotheoneobserved fainter magnitudes, some regions have significantly less stars
with UCAC4, with the main concentration being as expected thanexpectedfromthemodel.Theseregionsarelocatedspecif-
alongthedenseareasaddedtoasmallerGaiascanninglawfoot- ically around l = 200−250◦, b = 30−60◦ and l = 30−80◦,
print. b = −60;−30◦.Atmagnitudesfainterthan19,regionsallalong
theeclipticsufferfromthissmallernumberofsourcesduetothe
scanninglawandthefilteringofobjectswithatoolownumber
Quasars. Quasarsareessentialobjectsforvariousreasonsand
ofobservations.AlsoatG >16somediscrepanciesappearinthe
several tests verify that they have been correctly observed by
outerbulgeregions,whichmightbeduetoincompletenessofthe
Gaia and identified. The first test compares Gaia DR1 quasars
datawhenthefieldiscrowded(seeSect.4.3.1andFig.10).
with ground-based quasar compilations: GIQC (Andrei et al.
To estimate in more details the completeness in specific
2014), LQAC3 (Souchay et al. 2015) and SDSS DR10 (Pâris
fields, we compared histograms of star counts from Gaia DR1
et al. 2014) catalogues. It is a check for completeness, dupli-
andtheGOG18simulationasafunctionofmagnitude.Figure9
cation and magnitude consistency. While the quasars were also
affected by the duplicated sources issue (Sect. 3.2.1), the filter- showssuchhistogramsinsomeregionsofthegalacticplane,at
intermediate latitudes and at the Galactic poles. In the Galactic
ingseemstohaveremovedthemnicely.81%ofGIQC,53%of
plane (Fig. 9a) the star counts show a drop in the Gaia data at
LQAC3 and 11% of SDSS quasars are present in Gaia DR1, a
magnitudes brighter than in the model. This could be a priori
ratiothatreaches93%fortheLQAC3sourceswithamagnitude
duetoinadequateextinctionmodelormodeldensitylaws,orto
Bbrighterthan20.
incompletenessintheGaiadataatfaintmagnitudesduetounde-
tectedoromittedsources.Sincethebrightmagnitudecountsare
Galaxies. For galaxies, the cross-match has been done with fairly well fitted, the latter hypothesis is most probable. This is
SDSS DR12 (Alam et al. 2015) sources with a galaxy spectral alsopointedoutbycomparisonwithpreviouscatalogues.Inthe
classification.Thepropertiesofcross-matchedgalaxiesarecom- outerGalaxy,GOG18simulationisprobablyatooroughmodel
Articlenumber,page7of34
A&Aproofs:manuscriptno.GAIA-CS-CP-OPM-FA-072
1.0 1.0 1.0 1.0
60, 60, 0.8 60, 60, 0.8 60, 60, 0.8 60, 60, 0.8
0.6 0.6 0.6 0.6
0.4 0.4 0.4 0.4
120, 0, 0, 0, 240, -000.2.2RelDiff12 120, 0, 0, 0, 240, -000.2.2RelDiff13 120, 0, 0, 0, 240, -000.2.2RelDiff14 120, 0, 0, 0, 240, -000.2.2RelDiff15
-0.4 -0.4 -0.4 -0.4
-60, -60, -0.6 -60, -60, -0.6 -60, -60, -0.6 -60, -60, -0.6
-0.8 -0.8 -0.8 -0.8
1-1.0.0 -11.0.0 -11.0.0 -11.0.0
60, 60, 0.8 60, 60, 0.8 60, 60, 0.8 60, 60, 0.8
0.6 0.6 0.6 0.6
0.4 0.4 0.4 0.4
120, 0, 0, 0, 240, -000.2.2RelDiff16 120, 0, 0, 0, 240, -000.2.2RelDiff17 120, 0, 0, 0, 240, -000.2.2RelDiff18 120, 0, 0, 0, 240, -000.2.2RelDiff19
-0.4 -0.4 -0.4 -0.4
-60, -60, -0.6 -60, -60, -0.6 -60, -60, -0.6 -60, -60, -0.6
-0.8 -0.8 -0.8 -0.8
-1.0 -1.0 -1.0 -1.0
Fig.8.RelativestarcountdifferencesbetweenGaiaDR1andGOG18simulationindifferentmagnitudebins,from12<G<13to19<G<20
bystepofonemagnitude,ingalacticcoordinates.BesidetheprominentfeatureoftheMagellanicClouds(absentfromtheGalaxymodel),and
inadequaciesofthe3Dextinctionmodelinthegalacticplane,theGaiaincompletenessaroundtheeclipticplaneduetothescanninglawstarts
clearlytoappearfromG>16.
6.0 l=90 b=0 68..00 l=43-47 b=0 4.0 l=90 b=21
4.0 4.0
2.0 2.0 2.0
N_deg211ee8246800.....3400000 N_deg211ee468246800.......340000000 N_deg21e24680....30000
6.0 2.0
4.0
13 14 15 16 G 17 18 19 20 13 14 15 16 G 17 18 19 20 13 14 15 16 G 17 18 19 20
1e46820....30000 l=45 b=-45 3456789.......0000000 l=225 b=45 234567......000000 NGP
N_deg21e46820....20000 N_deg211ee456789200.......220000000 N_deg211ee345678900.......220000000
2.0
13 14 15 16 G 17 18 19 20 13 14 15 16 G 17 18 19 20 13 14 15 16 G 17 18 19 20
Fig.9.Starcountspersquaredegreeasafunctionofmagnitudeinseveraldirections.OpencircleslinkedwithredlinesareforGaiaDR1data,
filled blue diamonds are simulations from GOG18. Error bars represent the Poisson noise for one square degree field. The bottom row shows
regionsimpactedbythescanninglawandthefilteringofstarswithalownumberofobservations.
oftheGalacticstructures,ascanbeseeninthefieldsatlongitude scanninglaw,andbecausesourceswithasmallnumberofobser-
180◦ wherethesomesubstructuressuchastheMonocerosring vationshavebeenfilteredout.Thecompletenessisalsoreduced
or the anticentre overdensity might contribute. In Fig. 9b, the in the Galactic plane due to undetected or omitted sources in
field atlongitude43-47◦ andlatitude 0◦ isfor 2lines ofsights, crowdedregions.Thisisexpectedtobesolvedinfuturereleases
where the model (in blue) gives similar star counts for the two wherealargernumberofobservationswillbeavailable.
lines while the data (in red) do not. We believe that this is due
tovaryingextinction,whichisunderestimatedinthemodelfor
thesespecificfields. 4.3. SmallscalecompletenessofGaiaDR1
Over the whole sky, up to magnitude 18, there is a relative
differenceofafewpercent(fromlessthan3%atmagnitude12 4.3.1. Illustrationsofunder-observedregions
to 10% at magnitude 18). Between 18 and 19 the relative dif-
ferenceis15%.Intherange19to20,thedifferenceis25%on Empty regions due to the threshold on the number of observa-
the average. At high latitudes, and specifically at the Galactic tionsareillustratedinFig.10anearthegalacticcenter;regions
poles,theagreementbetweenthemodelandthedataisalsoquite under-scannedliketheseonesarenotfrequentandhavealimited
good. The regions where the Gaia data seem to suffer from in- area,below0.1squaredegree(seealsoGaiaCollaborationetal.
completenessarelocatedinthespecificregionsaroundl=225◦, 2016a, Sect. 6.2). The field shown in Fig. 10b near the bulge
b = 45◦ andl = 45◦,b = −45◦,mostprobablyrelatedtothefil- sufferedfromlimitedon-boardresources,whichcreatedholesin
tering of sources with a low number of observations. The data theskycoverage,asshownalsoforglobularclustersinFig.13.
are however probably complete up to G = 16 in those regions
(l = 225◦,b = 45◦),althoughtheincompletenesscouldalsooc-
curatbrightermagnitudesinsomeareas(atG = 14inl = 45◦, 4.3.2. Testswithrespecttoexternalcatalogues
b=−45◦).
These comparisons show that Gaia data have a distribution ThesmallscalecompletenessofGaiaDR1anditsvariationwith
over the sky and as a function of magnitude which is close to the sky stellar density has been tested in comparison with two
what is expected from a Galaxy model in most regions of the catalogues: Version 1 of the Hubble Space Telescope (HST)
sky.Howeveritpointstowardsanincompletenessatmagnitudes Source Catalogue (HSC, Whitmore et al. 2016) and a selection
fainter than 16 in some specific areas less observed due to the offieldsobservedbyOGLE(Udalskietal.2008).
Articlenumber,page8of34
F.Arenouetal.:GaiaDataRelease1–Cataloguevalidation
100 60
80 50
%) %)
mpleteness (in 4060 mpleteness (in 3040
Co Co 20
20
10
0
15 16 17 18 19 15 16 17 18 19
G_HST G_HST
Fig.11.GaiaDR1completeness(in%)versustheHubbleSourceCat-
alogueasafunctionofG magnitude.Thedottedlinescorrespondto
HST
the1σconfidenceinterval;a)inBaade’sWindow(l=1◦,b=−4◦);b)
forall-skyHSCsourcesobservedwiththeACSandtheF606Wfilter.
Fig. 10. Regions with under-densities in DR1: a) under-scanned field
nearl = 354◦,b = −3◦,size∼ 3squaredegrees;b)holescreatedby
lackofon-boardresourcesinanotherdensefieldnearl=330◦,b=3◦,
size∼200squarearcmin.
Hubble Source Catalogue. The HSC is a very non-uniform
catalogue based on deep pencil-beam HST observations made
usingawidevarietyofinstruments(WideFieldPlanetaryCam-
Fig. 14. Completeness of Gaia relative to HST in the area around
era 2 (WFPC2), Wide Field Camera 3 (WFC3) and the Wide
NGC5053featuringstellardensitiesunder1millionpersquaredegree.
FieldChanneloftheAdvancedCameraforSurveys(ACS)and
observing modes. The spatial resolution of Gaia is comparable
to that of Hubble and the HSC is therefore an excellent tool to
around the dwarf spheroidal galaxy Leo II (Lépine et al. 2011)
testthecompletenessofGaiaDR1onspecificsamplesofstars.
leadstoacompletenessatmagnitude20ofnearly100%,while
TocheckthecompletenessasafunctionofG,wecomputedan
a test for a high density area around the globular cluster NGC
approximateG-bandmagnitude from HSTF555Wand F814W
7078(Bellinietal.2014)leadstoacompletenessworsethanthe
magnitudes(G )usingtheoreticalcolour-colourrelationsde-
HST onepresentedFig.11.
rivedfollowingtheprocedureofJordietal.(2010).
Thefirsttestwasmadeinacrowdedfieldofonedegreera-
diusaroundBaade’sWindow.Nearly13000starswereconsid- HSTobservationsofGlobularClusters. Werundetailedcom-
ered,observedinboththeF555WandF814WHSTfilterswith pleteness tests within globular clusters using HST data specifi-
eitherWFPC2orWFC3. callyreducedforthestudyofthosecrowdedfields.Weused26
Thesecondtestwasmadeonsamplesofstarsobservedwith globular clusters for which HST photometry is available from
oneofthethreeHSTcameras,usingtheredfilterF814Wandei- thearchiveofSarajedinietal.(2007,seeTable1).Thedatafor
therF555WorF606W.Sourceswereselectedfollowingtherec- all GCs were acquired with the ACS and contain magnitudes
ommendations of Whitmore et al. (2016) to reduce the number in the bands F606W and F814W. The observations cover fields
of artefacts. Moreover, only stars with an absolute astrometric of3arcmin×3arcminsize.ForM4(NGC6121),databyBedin
correctionflaginHSTsettoyeshavebeenselected,leadingtoa etal.(2013),andMalavoltaetal.(2015)takenintheHSTproject
typicalabsoluteastrometricaccuracyofabout0.1(cid:48)(cid:48).Thesizeof GO-12911 in WFC3/UVIS filters were used. For this test, the
theresultingsamplesvariesfrom1600starsforACS-F555Wto photometric transformations HST bands to Gaia G-band were
nearly 120000 stars for ACS-F606W, going through 15-23000 adjustedforeachclustertofitasampleofbrightstarsinorderto
starsforthefourothersamples.ThecompletenessofGaiaobser- avoidissuesduetovariationsinmetallicityandextinction.
vationsforthesesamples,positiondifferencesandcolour-colour High quality relative positions and relative proper motions
relationshavebeentested. areavailablefortheseclusters.Whenartificialstarexperiments
The completeness results of both tests are presented in wereavailableintheoriginalHSTcatalogue(GCsmarkedwith
Fig. 11. In Baade’s Window, the completeness follows the ex- *inTable1),thecompletenessofHSTdatahasbeenevaluated
pectationsforDR1:inthisverydensearea,on-boardlimitations bycomparingthenumberofinputandrecoveredartificialstars
lead to a brighter effective magnitude limit. The “all-sky” re- in each spatial bin. We find the completeness of the HST data
sult(usinghere128000ACSstarswithF606W< 20mag)isat to be well above 90% and close to 100% in all cases for stars
firstsightmoresurprising,butinfactbrightsourceobservations brighterthanV =21,butfortheverycrowdedclusterNGC5139
withHSTarequiterareandaredonemainlyinverydenseareas (OmegaCen). The GCs are chosen to present different level of
(whichneedtheHSTresolution)suchasglobularclusters,which crowding down to G ∼ 22. In general, HST data cover the in-
also suffer from Gaia on-board limitations. We further checked nercoreoftheclusters,wherethestellardensitiesareabove106
thisinterpretationbyusingindividualHSTobservationsandim- stars per square degree in almost all regions (above 30 million
agesaroundafewpositions:thetestmadeforalowdensityarea inmanycases,andupto110millionstarspersquaredegreein
Articlenumber,page9of34
A&Aproofs:manuscriptno.GAIA-CS-CP-OPM-FA-072
Fig.12.CompletenessagainstdensityinthefieldofthreechosenGCs,indifferentmagnituderanges.FieldssuchasNGC1261haveamedianof
220observations,allowingforamuchbettercompletenessinthedenserregionsthanNGC6752(40observations).
Fig.13.StellardistributionforsixchosenGCs,colour-codedbynumberofGobservationforeachstar.Toprow:examplesofholescausedby
limitedon-boardresourcesorbrightstars.Bottomrow:insomeregionspatternsarevisiblecorrespondingtostripeswherenostarshadasufficient
numberofobservations.
the core of NGC 104/47 Tuc). In a few cases, lower densities and OGLE-IV LMC (Soszyn´ski et al. 2012) surveys. A G-
arereachedintheexternalregions.WethereforeexpectGaiato bandmagnitudewascomputedfromOGLEV andImagnitudes
be very severely incomplete in most of the regions studied in (G ) using an empirical relation derived from the matched
OGLE
this test. The HST magnitudes were converted to GaiaG mag- Gaia/OGLEsources(tworelationswerederived,oneforOGLE-
nitudesusingthesametransformationsaspreviouslybetweenG IIIandoneforOGLE-IVduetotheirdifferentfilters).Thestellar
andF814W,F606WbutontheVegaphotometricsystem. densitieswereestimatedfromtheOGLEdatathemselves,there-
For each GC, the total density of stars in square bins of foretheyarecertainlyslightlyunder-estimated.Ascanbeseen
0.008deg = 0.5arcmin was evaluated, then in each bin we in Fig. 15, the completeness is not only dependent on the sky
countedthenumberofstarspresentintheHSTphotometryand density,butalsoontheskyposition,linkedtotheGaiascanning
intheGaiaDR1,bysliceinmagnitude. law,aswesawabove.Inthebulgefields,thecompletenessmay
ThecompletenessofGaiaDR1isshowninFig.12forthree show a drop aroundG=15 (as seen in Fig. 15b, confirming the
clusters,asafunctionofthestellardensityobservedintheHST featureofFig.11a).Thisisduetothefactthatthereddeststars
data.Differentcrowdedregionspresentdifferentdegreesofcom- havenotbeenkeptinGaiaDR1(becauseoffilteringatcalibra-
pleteness, depending on the number of observations in that re- tion level) and those missing stars correspond to the reddened
gion. In addition, holes are found around bright stars (typically redgiantbranchofthebulge(Fig.15c).
forG <11−12mag),andentirestripesaremissing,asillustrated
inFig.13.
In less crowded regions, such as in the field around 4.4. Completenessandangularresolution
NGC5053wherestellardensitiesareunder1millionpersquare
Althoughtherearenodoubtsabouttheexcellent,spatialangular
degree,thecompletenessisveryhigh,asshowninFig.14.
resolutionofGaia3,theeffectiveangularseparationinGaiaDR1
canbequestioned,e.g.duetopossiblecross-matchproblems.
OGLE catalogues. To further test the variation of the com-
pleteness with sky density, we looked at the completeness ver-
sus OGLE data using a few fields in the OGLE-III Disk (Szy- 3 e.g.PlutoandCharoncouldeasilybeseparatedwitha0.36"along-
man´ski et al. 2010), OGLE-III Bulge (Szyman´ski et al. 2011) scanseparation,seehttp://www.cosmos.esa.int/web/gaia/iow_20160121
Articlenumber,page10of34